In order to facilitate the routing of data in a network, in particular when this network comprises a large number of nodes, it is known to organize the nodes of this network into groupings (also called “clusters”).
A node of each cluster, termed the master node, is chosen to ensure the organization and maintenance of the cluster of nodes of which it forms part, in particular by serving as gateway between this grouping and the other nodes of the network. For this purpose, the master nodes of the clusters communicate with one another for routing data from one node of one cluster to another node of another cluster, by serving as intermediary between these nodes.
It will be noted that the organization of the nodes of a network into clusters is particularly necessary in the case of an ad hoc network, that is to say a network not comprising any predetermined infrastructure, in which furthermore there is only a limited bandwidth and in which the nodes are potentially mobile.
The nodes of an ad hoc network being potentially mobile, it is sometimes necessary to reorganize the clusters of nodes. Thus, it is preferable to have available a fast and effective method of organizing the nodes of the network into clusters of nodes.
A fast and effective method such as this is particularly necessary in a large-scale ad hoc network, that is to say an ad hoc network comprising a large number of nodes, for example more than a hundred nodes, and whose nodes are particularly mobile.
A method of organizing nodes of a network by grouping nodes is already known from the prior art, in which at least one node, termed the master node, is chosen to serve as gateway between a grouping of nodes and other nodes of the network, the method comprising:                a step of calculating, for each node, a numerical value, termed the weight, representing the aptness of this node to be the master node of a cluster of nodes,        a step of electing, from among the nodes, a master node of a cluster, such that the weight of this master node is greater than or equal to the weights of each other node of the cluster comprising this master node.        
In general, the weight of a node is calculated by taking account of parameters corresponding to properties inherent to this node, such as, for example, its mobility, its transmission power, the extent of its coverage zone, the power of its battery, its connectivity, etc.
The weight of a node can also be calculated by taking account of parameters corresponding to structural properties of the network, such as, for example, the number of nodes adjacent to the node whose weight is calculated.
In accordance with this conventional method for organizing nodes, each node of the network transmits an information cue containing its weight towards the nodes adjacent to it, generally by flood routing. The node chosen to be the master node is then the node whose weight is higher than the weights of all the nodes adjacent to it.
It will be noted that such a method is generally expensive in resources, and that it generally makes it possible to form only clusters of nodes of small sizes. Specifically, although it is possible to organize the nodes into clusters of larger sizes, each node must, in this case, transmit the information cue relating to its weight to a large number of nodes, thereby considerably increasing the costs in terms of resources and system time.